METHODEO

The major challenge for video-surveillance is the huge amount of data that are produced, even by medium size systems. As an event arise, time spent by users to retrieve the data of interest is therefore very long, and is requesting an strong involvement of them that has a impact in terms of cost. The delay for delivering the information of interest (the proof image) is one of the critical issues (the image quality is another one), with a direct impact for the security of the citizens and the perceived efficiency of video-surveillance systems. Video analytics algorithms have progressed along the time and can be envisaged for helping the retrieval of information in large video-surveillance systems, whereas this task is still done manually (as an example for RATP and SNCF public transports operators). For enabling the use of such algorithms it is therefore mandatory to be able to predict with a good precision the expected performances of such algorithms, when they will be applied on real video-surveillance data. This issue is very critical for video-analytics as the context differs from the assessment of big software systems due to the very large variability of parameters. It is also difficult or even unrealistic to do an exhaustive test on a database covering all situations and contexts. The main objective of METHODEO is to establish an evaluation methodology for the video-analytic algorithms, especially for the exploitation of video-surveillance recordings, for being able to anticipate their behaviour on different types of data bases and contexts. After a study and assessment of the pre-existing methodologies, with consideration of the user’s requirements and the realistic contexts, the METHODEO methodology will be studied and developed. The tracks to be followed by the project and its technical orientations will be decided by the Strategic Committee, chaired by the coordinator, with a representative person from each of the partners, linked to the Steering Committee that will involve three users of METHODEO: • Steven Aimé - STSI/CTSI • Marie-Hélène Bonneau - UIC • Jean-François Sulzer - Thales Security Solutions and Services SBL Protection Systems